Mark Graves Leveraging Existing DBMS Storage for XML DBMS

This presentation is Copyright 2001, 2002 by Mark Graves and contains material Copyright 2002 by Prentice Hall PTR. All rights reserved.

Agenda DBMS Architecture External Interface Data Model Storage Systems –Overview –Fine-grained RDBMS storage –Coarse-grained RDBMS storage –Medium-grained RDBMS storage

XML DBMS Create a DBMS to capture XML Access of document & elements Should support: –Storage –Querying –Editing

DBMS Architecture

External Interface User Interface -- HTML or Java URLs -- command access Java API -- used by servlet Command-line interface XML -- taglib, SOAP

Data Models Type Constructors Operations Constraints Examples: relational, Entity- relational, object

XML Data Model -- Types Document has one name and one (root) element. Element has –type name (which is a string), –collection of attributes, and –ordered collection of (interspersed) character data and elements. Attribute has a name and a value (both strings). Character data has a value (a string).

XML Data Model -- Constraints Each document name may occur only once. (Thus, the document names are unique and may be queried.) All elements other than the document element have an element node as a parent. The document element has no parent. (Thus, the elements form a tree.) No attribute name may appear more than once in an element.

XML Data Model -- Operations Add and Delete Retrieve Replace Search

Operations -- Add and Delete Add a document to the database. Delete a document from the database. Add an element to a specific location in the document. Delete an element from a specific location in the document.  Add an attribute to an element.  Delete an attribute from an element.

Operations -- Retrieve Retrieve a document from the database given its name. Retrieve an element from a specific location in the document. Retrieve all the elements and character data from a document in document order (in effect, regenerate the document). Retrieve an attribute from an element given its name.  Retrieve the nth child of an element.  Retrieve all children of an element.  Retrieve the text of the character data.  Retrieve the parent element of the character data.

Operations -- Replace  Replace an element at a specific location with another element or character data. Replace character data at a specific location with other character data or elements.  Replace the value of an attribute in an element given its name.  Set the text of the character data.

Operations -- Search  Search for all documents in the database given a particular set of constraints.  Search for all elements in a document that satisfy a particular set of constraints. Search the document for character data that matches a particular set of constraints (such as matching a string).  Element type name equals (or does not equal) some value.  Attribute name equals (or does not equal) some value.  Character data equals (or does not equal) some value.  Element has a specified number of children (or less than, or greater than, or not equal to).  Character data contains a specified string as a substring.  Query constraint consists of two query constraints that must both be true (or either be true).  Query constraint consists of one query constraint that must not be true.

Storage System (Internal Interface) Native store Object-oriented Complex flat-file Relational DBMS

Leveraged Storage Systems

RDBMS Implementation Use a Relational DBMS to store XML documents Strategies –fine-grained -- store every piece of data separately (completely parsed) –coarse-grained -- store entire document together (no parsing) –medium-grained -- store some elements in coarse-grained storage, other in fine- grained storage (partial parsing)

Fine-grained Storage Approach: Completely parse data and store each element, attribute, and character data value in a relational table. Design –Conceptual Schema –Logical Schema (unnormalized & normalized) –Physical Schema Implementation (Java)

Conceptual Schema

Fine-grained Logical Schema Document(name DOC_NAME, root ELEMENT) Element(doc DOCUMENT, parent ELEMENT, tag ELE_NAME) Attribute(doc DOCUMENT, element ELEMENT, name ATTR_NAME, value ATTR_VALUE) CharData(doc DOCUMENT, element ELEMENT, value CDATA) Child(doc DOCUMENT, element ELEMENT, index NUMBER, child_class CHILD_CLASS, child CHILD_NODE)

Fine-grained Logical Schema

Fine-grained Physical Schema

Fine-grained Commands Retrieve a document (with or without XML header) Store a document Delete a document List documents in database

Fine-grained Implementation

Coarse-grained Storage Approach: Store each document in its entirety Logical Schema: –Document (name STRING, body TEXT) Physical Schema:

Medium-grained Storage Use both fine-grained (parsed) and coarse-grained (unparsed) storage as appropriate within a document Slice points Multiple slice points Specifying slice points –element type name –element type name & attributes

Dictionary Example

Dictionary Example

Medium-grained Physical Schema